Scalable Memory Allocation for Parallel Algorithms
In a multithreaded C/C++ program, using standard non-threaded allocators, memory allocation can become a bottleneck. That is firstly caused by thread competition for a lock on a shared global heap, and secondly for caching effects. Programs that run this way are not scalable and may slow down as the number of cores increases. Scalable memory allocators such as Intel’s TBB allocators, FreeBSD’s jemalloc and Google’s TCMalloc solve this problem by providing various optimizations such as per-CPU caches, thread-private heaps, sized deletes and fast/slow path improvements. You can easily gain a 20-30% performance improvement for parallel sections and even 4X in extreme cases by simply relinking with a scalable memory allocator. This webinar will tell you all about these allocators, with a live session running some benchmarks at the end. Materials presented during the live session are available on GitHub (https://github.com/arminms/scalable_a....
_______________________________________________
This webinar was presented by Armin Sobhani (SHARCNET) on March 17, 2021, as a part of a series of regular biweekly General Interest webinars ran by SHARCNET. The webinars cover different high performance computing (HPC) topics, are approximately 45 minutes in length, and are delivered by experts in the relevant fields. Further details can be found on this web page: https://www.sharcnet.ca/help/index.ph... . Subscribe to our twitter account (@SHARCNET) to stay updated about our upcoming webinars.
SHARCNET is a consortium of 19 Canadian academic institutions who share a network of high performance computers (http://www.sharcnet.ca). SHARCNET is a part of Compute Ontario (http://computeontario.ca/) and Compute Canada (https://computecanada.ca).
On this page of the site you can watch the video online Scalable Memory Allocation for Parallel Algorithms with a duration of hours minute second in good quality, which was uploaded by the user Sharcnet HPC 17 March 2021, share the link with friends and acquaintances, this video has already been watched 406 times on youtube and it was liked by 5 viewers. Enjoy your viewing!
